Temporal constraint relationship for use in an image segregation for frames of a video
View Patent ↗A soft, weighted constraint imposed upon image locations temporally spaced in frames of a video, can be used to provide a more accurate segregation of an image into intrinsic material reflectance and illumination components. The constraint is arranged to constrain all color band variations between the image locations into one integral constraining relationship.
1. An automated, computerized method for processing multiple image files of a video, comprising the steps of:
providing at least two image files, each depicting an image having image locations, in a computer memory;
selecting image locations i from a first image file k and j from a second image file k−n, k−n being temporally spaced n frames from image file k in the video;
setting a temporal constraint between the image locations according to a log color space equation:
w n ij *( Mi−Mj )= w n ij *(α ij V ij +(1−α ij ) B ij ),
wherein Mi and Mj are log color space values for the material reflectance component of the image locations i and j, respectively, V ij is a difference vector to provide a gradient magnitude defined by V ij =I i −I j , where I i and I j are log color space values corresponding to color values recorded in the image files, for the color and intensity of the selected image locations i and j, respectively, B ij is a component of V ij not aligned with illumination change between the selected image locations and is a vector defined by B ij =V ij −(V ij ·N), where N is a BIDR normal, α ij is a scalar value based upon a smooth scalar function α that varies from 0 to 1, as a function of a degree of change expressed by I i −I j corresponding to one of an illumination change or a material reflectance change, and w n ij is a temporal constraint confidence weight calculated as w n ij =s n *w ij , wherein w ij is a constraint weight and s n is a scaling factor determined as a function of the n frames between the image file k and the image file k−n frame of the video; and
solving the log color space equation to identify an intrinsic component of the selected image locations.
2. The method of claim 1 wherein the scaling factor s n is calculated according to an equation:
s n =T w *exp(− n 2 /2σ t 2 ),
wherein T w is a pre-set temporal weight and σ t is a temporal sigma calculated as a function of n.
3. A device which comprises:
a computer; and
a memory storing at least two image files of a video, each depicting an image having image locations;
the computer is arranged and configured to execute a routine to select image locations i from a first image file k and j from a second image file k−n, k−n being temporally spaced n frames from image file k in the video, set a temporal constraint between the image locations according to a log color space equation:
w n ij *( Mi−Mj )= w n ij *(α ij V ij +(1−α ij ) B ij ),
wherein Mi and Mj are log color space values for the material reflectance component of the image locations i and j, respectively, V ij is a difference vector to provide a gradient magnitude defined by V ij =I i −I j , where I i and I j are log color space values corresponding to color values recorded in the image files, for the color and intensity of the selected image locations i and j, respectively, B ij is a component of V ij not aligned with illumination change between the selected image locations and is a vector defined by B ij =V ij −(V ij ·N), where N is a BIDR normal, α ij is a scalar value based upon a smooth scalar function α that varies from 0 to 1, as a function of a degree of change expressed by I i −I j corresponding to one of an illumination change or a material reflectance change, and w n ij is a temporal constraint confidence weight calculated as w n ij =s n *w ij , wherein w ij is a constraint weight and s n is a scaling factor determined as a function of the n frames between the image file k and the image file k−n frame of the video and
solve the log color space equation to identify an intrinsic component of the selected image locations.
4. A computer program product, disposed on a non-transitory computer readable media, the product including computer executable process steps operable to control a computer to: receive at least two image files of a video, each depicting an image having image locations, select image locations i from a first image file k and j from a second image file k−n, k−n being temporally spaced n frames from image file k in the video, set a temporal constraint between the image locations according to a log color space equation:
w n ij *( Mi−Mj )= w n ij *(α ij V ij +(1−α ij ) B ij ),
wherein Mi and Mj are log color space values for the material reflectance component of the image locations i and j, respectively, V ij is a difference vector to provide a gradient magnitude defined by V ij =I i −I j , where I i and I j are log color space values corresponding to color values recorded in the image files, for the color and intensity of the selected image locations i and j, respectively, B ij is a component of V ij not aligned with illumination change between the selected image locations and is a vector defined by B ij =V ij −(V ij ·N), where N is a BIDR normal, α ij is a scalar value based upon a smooth scalar function α that varies from 0 to 1, as a function of a degree of change expressed by I i −I j corresponding to one of an illumination change or a material reflectance change, and w n ij is a temporal constraint confidence weight calculated as w n ij =s n *w ij , wherein w ij is a constraint weight and s n is a scaling factor determined as a function of the n frames between the image file k and the image file k−n frame of the video and
solve the log color space equation to identify an intrinsic component of the selected image locations.
5. The computer program product of claim 4 wherein the scaling factor s n is calculated according to an equation:
s ij =T w *exp(− n 2 /2σ t 2 ),
wherein T w is a pre-set temporal weight and σ t is a temporal sigma calculated as a function of n.